For those wondering how bad the issue is, check out this[1] screenshot of a diff between the TLS/non-TLS html from www.example.com.
Here's "corrupted" UTF-8 characters expressed as bits. In this example you can see clearly what's happening. The top byte is correct and the second one is corrupted.
01100111 # g
01101111 # o
01110000 # p
01111000 # x
01101100 # l
01100100 # d
01100101 # e
01101101 # m
Here's "corrupted" UTF-8 characters expressed as bits. In this example you can see clearly what's happening. The top byte is correct and the second one is corrupted.
1. https://i.imgur.com/qIMnTTe.png